def __call__(self, filesets): """ The algorithm itself """ # Update run list self.getNewRuns() # Do per fileset work, abandon fileset processing on exception for fileset in filesets: ds = fileset.name try: # Do per run work watchCompleteFiles = [] for watch in self.watchedRuns: # Ensure watcher has dataset listed watch.addDatasetOfInterest(ds) # Query DBS to find all blocks for this run / dataset (files, blocks, fileInfoMap) = \ self.dbsHelper.getFileInfo(watch.run, ds) # Now determine all required parent blocks parentBlocks = set() if fileset.requireParents: parentDs = self.dbsHelper.getParentDataset(ds) parentBlocks = self.dbsHelper.getBlockInfo( watch.run, parentDs) # Final list of all required blocks allBlocks = blocks[:] allBlocks.update(parentBlocks) # Find all sites where all blocks are complete sites = self.phedexHelper.getCompleteSites(blocks) # Get sites with newly completed transfers newSites = watch.getNewSites(ds, sites) if len(newSites) > 0: # Add the files for these blocks to the fileset for file in fileInfoMap: fi = fileInfoMap[file] # First add parent file if fileset.requireParents: parentFile = File(lfn=fi["file.parent"]) parentFile.save() parentFile.setLocation(newSites) # Add actual file fileToAdd = File(lfn=file, size=fi["file.size"], events=fi["file.events"], run=watch.run, umi=fi["file.lumi"]) if not fileToAdd.exists( ) and fileset.requireParents: fileToAdd.addParent(fi["file.parent"]) # Add new locations but don't persist immediately fileToAdd.setLocations(newSites, immediateSave=False) # Add the file to the new file list fileset.addFile(fileToAdd) # Add the site info to the watcher list watchCompleteFiles.append([watch, ds, newSites]) # Commit the fileset fileset.commit() # Add the watched runs for a in watchCompleteFiles: a[0].addCompletedNodes(a[1], a[2]) except: # Reset the watch list so we re-evaluate next call watchCompleteFiles = [] # Purge old runs self.purgeWatchedRuns()
def __call__(self, filesets): """ The algorithm itself """ # Update run list self.getNewRuns() # Do per fileset work, abandon fileset processing on exception for fileset in filesets: ds = fileset.name try: # Do per run work watchCompleteFiles = [] for watch in self.watchedRuns: # Ensure watcher has dataset listed watch.addDatasetOfInterest(ds) # Query DBS to find all blocks for this run / dataset (files, blocks, fileInfoMap) = \ self.dbsHelper.getFileInfo(watch.run, ds) # Now determine all required parent blocks parentBlocks = set() if fileset.requireParents: parentDs = self.dbsHelper.getParentDataset(ds) parentBlocks = self.dbsHelper.getBlockInfo(watch.run, parentDs) # Final list of all required blocks allBlocks = blocks[:] allBlocks.update(parentBlocks) # Find all sites where all blocks are complete sites = self.phedexHelper.getCompleteSites(blocks) # Get sites with newly completed transfers newSites = watch.getNewSites(ds, sites) if len(newSites) > 0: # Add the files for these blocks to the fileset for file in fileInfoMap: fi = fileInfoMap[file] # First add parent file if fileset.requireParents: parentFile = File(lfn=fi["file.parent"]) parentFile.save() parentFile.setLocation(newSites) # Add actual file fileToAdd = File(lfn=file, size=fi["file.size"], events=fi["file.events"], run=watch.run, umi=fi["file.lumi"]) if not fileToAdd.exists() and fileset.requireParents: fileToAdd.addParent(fi["file.parent"]) # Add new locations but don't persist immediately fileToAdd.setLocations(newSites, immediateSave=False) # Add the file to the new file list fileset.addFile(fileToAdd) # Add the site info to the watcher list watchCompleteFiles.append([watch, ds, newSites]) # Commit the fileset fileset.commit() # Add the watched runs for a in watchCompleteFiles: a[0].addCompletedNodes(a[1], a[2]) except: # Reset the watch list so we re-evaluate next call watchCompleteFiles = [] # Purge old runs self.purgeWatchedRuns()